<> preface ：
Tensorflow yes Google Machine learning open source artifact launched , yes Python With good language support , support CPU,GPU and Google
TPU Other hardware , And has a variety of models and algorithms . at present ,Tensorflow It has been widely used in text processing , Many machine learning and deep learning fields such as speech recognition and image recognition .
<> Basic framework
It is divided into three layers ： application layer , Interface layer and core layer
application layer ：
It provides a training library related to machine learning , Prediction library and for Python,C++ and Java And so on , be similar to web Front end of system , It mainly realizes the construction of calculation graph .
Interface layer ：
yes Tensorflow Packaging of functional modules , It is convenient for other language platforms to call .
Core layer ：
The most important part , Including equipment layer , network layer , Data operation layer and graph calculation layer , Perform calculation of application layer .
1. Equipment layer
include Tensorflow Implementation on different hardware devices , Main support CPU,GPU and Mobile Other equipment , Realize the conversion of calculation commands on different hardware devices , Provide a unified interface to the upper layer , Realize the cross platform function of the program .
2. network layer
The network layer mainly includes RPC and RDMA communication protocol , Realize data transmission and update between different devices , These protocols will be used in distributed computing .
3. Data operation layer
with tensor For processing objects , realization tensor Various operations and calculations .
4. Figure calculation layer
Including the implementation of distributed computing graph and local computing graph , Implementation diagram creation , compile , Optimization and implementation, etc .
Technology
Daily Recommendation